Global Guerrillas

Journal: AI Facial Recognition, Troll Shock Troops, and Online Polarization

Some items of interest:

• Singapore: "Changi (airport) will be using facial recognition technology to offer self-service options at check-in, bag drop, immigration and boarding" What happens over the long term? Automated systems like this will seek to partner with companies like Facebook.  Why?  Facebook will have, if they don't have already, the best facial recognition AIs available.  Why?  They simply have the biggest and best dataset available to build it with, by a wide margin.  They also can do it on a global scale (sans China/Russia) and they have strong vectors for expansion/growth into more robust methods of identity verification. 
  
  
• Oxford University:  Where are the social media trolls and bots?  A new working paper from Oxford (Troops, Trolls and Troublemakers: A Global Inventory of Organized Social Media Manipulation) assembled a global inventory of cyber shock troops used by governments, militaries, and political parties to manipulate social media.  Unsurprisingly, given that the US pioneered this technology and excels at all things marketing and media, the US leads the world in this new arms race:

Cyber Shock Troops

• Princeton University:  What led to the US political factionalization?   One of the big reasons is that the US news media lost its ability to set the context for the news (the facts).   Traditionally, this context was set through the emotional and moral conjugation of the facts.  For example, "the President's statement" is transformed through emotional conjugation into "the President's outrageous statement".  The advent of social networking formally ended the media's monopoly on setting the emotional context for the news (although the division began earlier with Fox, Drudge, etc.). The US emotional context is now deeply divided.  Here's a Princeton study that mapped the use of emotional/moral words online (pic below).  It showed the US is deeply divided, by the emotional and moral context that frames the news (a far more interesting finding that blaming "fake news").   It also shows there is little crossover between the two groups, likely due to the way social networking self-reinforces the context and controls the people within it.  What?  People don't just read/hear the news anymore -- they retweet it, like it, add emotionally conjugated commentary to it, and pressure their friends with it now.  

The use of moral words

Welcome to Facebook's Censorship and Surveillance Platform

I recently ran into a European counter-terrorism expert who complained that it was getting very difficult to build a fake profile on Facebook.  Every time his team tried to set up a fake profile, it was shut down in less than 24 hours.  Here's why he ran into problems.

• Facebook has an initiative to prevent the creation fake accounts (something Facebook strangely calls recidivism).  
• This initiative is a small part of a larger overall effort being undertaken by Facebook, Google and others, to become what can best be described as fully functional global censorship and surveillance systems.  I know that people have been concerned about this for a while, but it's not speculation anymore folks.  It's here.  
• The surprising thing to me?  The US and nearly all of the governments of the world (outside of China and Russia) are pushing them to do it.  

A global censorship and surveillance platform

Here are some of the aspects of these efforts:

• AIs that can identify violent imagery and extremist symbols in videos and pictures and rapidly delete them -- or better yet, block their upload or shut down a livestream as soon as they show up.  For example:  a live broadcast during a terrorist attack or murder (both happened).  
• Routine censorship and surveillance.  For example:  Facebook has ~7,500 (largely low paid subcontractors) reading posts and (private) messages to find and delete content they deem objectionable and ban the people who post it.  However, these folks are just temporary employees.  The real goal is to build AIs that can read posts and messages to ID objectionable content to do what the human team (above) is already doing, but on a global scale.  
• A complete social graph.  A real-time census of every living person in the world (outside of China and Russia).  One that knows all about you, whether or not you are on Facebook/Google/etc.  These companies are already close to this goal in Europe and the US, and at 2 billion daily users (Facebook and Android), so it won't be long before they expand that to the rest of the world.  

Where is this Headed?

The decline of the US security framework at home and abroad, growing political and economic instability and widespread distrust/illegitimacy will make an expansion of this platform inevitable.  Let's look at this expansion from a couple of angles:  

• Already, the social networks are replacing the media as the gatekeepers and the shapers of national and global public opinion.  It's clear that the media can't play this role anymore, they are outgunned. To wit: millions watch TV news while billions get their news from Facebook.  How will they replace the media?  They will use AIs to subtly block, blur or bury fake or objectionable information and conversations while promoting those they approve of.  This process will become extremely apparent during the next presidential campaign.  It also suggests that we will see candidates from within these companies running for office in many countries, and given their edge in using these platforms: win.
• AI's built using real time and detailed social graph information could become better at detecting violent behavior far sooner and better than human analysts.  Simply, it may not matter if the attackers were using Facebook or Google, the ripples from their actions on adjacent social networks might be more than enough to detect them.  Pushing this forward even further, as the data flows and the depth of the information increases, how far down in the stack of violence could these AIs prove to be effective?  Down to domestic murders, abuse, and rape?    
• A global ID.  Simply, Facebook is getting close to being able to create a global ID for everyone on the planet (sans China/Russia).  It's not a bit of paper or something you put in your wallet.  It'll be passive.  It'll replace your passport and driver's license. If you can be seen by a camera, you will be known.  

Sincerely,

John Robb

 

Robotic Systems Disruption in Practice

The Ukrainian SBU now believes that the destruction of arms depot at Balakliya in March that did a billion dollars in damage was carried out by a small drone armed with a thermite grenade.  That's an ROI (Return on Investment) of $500,000 for every $1 invested (not bad relative to earlier comparatives).

This is a spot on demonstration of what I wrote back in February of 2016.  Here's a recap of what I wrote in that article:  

From a mechanical perspective, consumer drones aren't that impressive:

• ~1-2 pound payload
• ~20 min flight time 
• 20-40 miles per hour flight speed

However, these drones are smart and the smarter the drone is, the better it can mimic the performance of the much more expensive precision guided munition (PGM).  For example:

• Drones can fly themselves.  They can take-off, fly enroute, and land autonomously.
• Drones can precisely navigate a course based on the GPS waypoints you designate.
• Drones can now (a recent development) use digital cameras to find, track, and follow objects.  Some can even land on objects they find based on a description of that object.  

Even this basic capability is more than enough to turn a basic drone into an extremely dangerous first strike weapon against fragile/explosive targets. Here's a scenario that pits ten drones against a major airport:  

1. Ten drones would take off autonomously in 1 minute intervals.
2. Each would follow a GPS flightpath to a preselected portion of an airport.
3. Upon arrival, a digital camera would identify the nearest wing of an aircraft.
4. The drone would land itself in the middle of that wing.
5. A pound of thermite in the payload would ignite upon landing.  
6. The thermite would burn through the wing, igniting the fuel inside...
7. Most of the airport and nearly all of the planes on the tarmac are destroyed.

Here are the takeaways:  

• Even the simple robotic platforms of today can be extremely effective as weapons.  At current rates of improvement in machine intelligence, the situation will get much more interesting very, very soon. 
• It's possible to creatively trade inexpensive machine smarts for expensive mechanical performance.  
• We need to figure this out before the bad guys do.  However, truly figuring this out requires a deep insight into the dynamics driving this forward.  

Sincerely,

John Robb

PS:  The Balakliya follows five earlier attacks on warehouses and even this facility since December 2016 using the same technique.  They are getting better.

 

How to turn the AI Economy into wealth, income, and pensions for millions

Six years ago, I floated a business plan for building a new type of company (an open source venture). An open source venture is structured in a way that makes it possible for millions of worker/owners to collaborate on the construction of extremely valuable databases. That initial attempt wasn’t successful, with good reason: it was too early. The databases that could be built weren’t as valuable as they needed to be to support millions of contributors and the system needed to manage decentralized ownership (the blockchain) was still in its infancy.

Things have changed. The rapid rise of AI (more specifically, deep learning/ConvNet), has turned massive, human labelled/curated, databases into some of the most valuable information on earth. Second, the blockchain is now mature enough to support the infrastructure needed to make it possible.

Here’s some more detail. Building AIs requires three things:

• Scalable models of AI construction with high capacity (we have that now deep learning/ConvNet now).
• Lots of processing power (to train AIs). Moore’s law cracked the initial barrier to using these methods. Cloud based training systems filled with GPUs is making this less expensive.
• Big data sets (specifically, human curated data) are used by the AI’s to learn.

You should already see the how well suited open source ventures filled with millions of worker owners fits into this process. For example:

• A recent study by Google showed that the size of the training database correlated to the quality of the AI. The bigger the better. That’s a bit of a problem for the big tech companies. Currently, most of the of the databases currently available were built by very low cost workers for a fraction of a penny a label, using places like Amazon’s Mechanical Turk. That makes them clunky to build and difficult to scale. Further, the labelled, human curated data they do have (on Facebook, Google, etc.), while voluminous, isn’t complete enough to do high quality training. An open source venture fills the void. It could enlist millions of worker/owners to build, maintain and extend the data needed would be very successful here, creating some of the most valuable AIs in the world.
• The processing power needed to train these AIs, as they become more sophisticated, could outstrip current corporate capabilities (of even Google). This suggests that training systems that break apart the computation tasks among tens of millions of participants (cell phone/desktop) could surmount this barrier. In practice, it would be a combination of folding@home and bitcoin mining. Already, deep learning AIs can “run” on cellphones. Configuring them to contribute to the training load and/or actively gather data for database construction isn’t that much of a stretch.
• The people participating (earning equity instead of fee for service for their contributions) could become very, very large. They would be familiar with the platform and become increasingly sophisticated at finding amazing ways to capture new data and new AIs to build. They would also be a built in audience of early adopters for new AIs, pushing them across the chasm into the mainstream and radically improving their operation due to the sophistication of their interaction with them. Over time, a platform like this could become the source of many (if not most) of the best AIs ever built. A source of immense wealth (seen as dividends) for hundreds of millions of owners, earning equity with each contribution.

Worth thinking about.

John Robb

PS: Given our experience with bitcoin, this isn’t impossible to do. It’s also a much better future than a world that one based on Turking day laborers.

JOURNAL: Some actionable information on rising sea levels

It's a pretty good bet we're going to see a rise in sea levels (we already have).   Here's a good report by the Union of Concerned Scientists with some much needed detail on how it will impact us.

How it will do damage:  Rising sea levels will have its greatest impact in terms of chronic tidal flooding.  Basically, every couple of years, the first floor of your house is filled with seawater destroying everything.

Where will chronic flooding occur?  Three maps.  The first one is an intermediate scenario for 2035 (less than 20 years out) and second is one for 2060.

Expected by 2035

Note that the impact by 2035 is mostly in four areas (DE, FL, LA, and NC).    By 2060, that damage deepens and expands considerably.  It turns almost all of the eastern (including NYC and Boston) and southern coasts into chronic tidal flood zones. 

Expected by 2060

Here's the last.  2100.  Note that even by 2100 chronic tidal flooding is largely a eastern/southern coast problem.  The West coast does relatively well.  

Expected by 2100

Hope this actionable information helps you make better decisions (or at a minimum, let's you effectively process the news this situation will produce).  

John Robb